In the past decade, the growth of Artificial Intelligence within our society has been rapid and seamless. It seems as though until only very recently, with the introduction of the GDPR and scandals such as that of Cambridge Analytica, we have allowed with open arms the influx of AI in our daily lives, barely even giving a thought to the limitless negative repercussions that we consider today. As a consequence of this, there has been almost no slowing of the research and development of these technologies, which now already wields an unprecedented amount of power over us.
Wanting to be able to understand and communicate with AI technologies as best we can, we push towards creating systems that communicate more and more as humans do. Naturally, since we are able to best convey our messages to each other (i.e. other humans), we now attempt to make AI systems as human-like as possible, to be able to maintain such a high level of correspondence. There is danger however, in pursuing the creation of AI that perfectly resembles a human. We believe that if such a robot AI would exist, human-to-human contact would be massively diminished, as human-to-robot contact would be far preferred, due to such a robot being created to simply appease mankind’s thoughts and beliefs, instead of challenge them. Furthermore, the inherent biases that exist in the creation of AI could snowball to be disastrous in the case of humanoid robots that physically exist in the real world, causing real danger. These subjects raise concerns that are not widely considered when speaking of AI today.
Not only are AI systems becoming more and more humanlike, but humans are going more and more in the opposite direction, that is, becoming AI-like. With the introduction of Elon Musk’s Neuralink, the interface that would connect human brains and computers to facilitate superintelligence in humans, the jump between the current human and the perfect AI suddenly becomes all too real. We similarly deem this influx of AI thinking capacity to the human brain an alarming step in the convergence of humans and AI. Neuralink being introduced into society in a commercial manner would undoubtedly only widen the equality gap, as those who would possess it would be far superior in thinking capacity to others. Furthermore, the inclusion of third-party AI capacities in the human brain is a far too frightening thought, given the lengths that companies go to in advertisement and publicity in the modern day and age. It is imperative that we consider such issues far before the technology reaches the point of realization, as there can only be more problems that even we can perceive at this moment.
AI Becoming More Human-Like
The classical depictions of human-like artificial intelligence in popular media in forms such as Arnold Schwarzenegger’s character in The Terminator, or Sonny from iRobot, show us the supposed ‘extremes’ of human-like AI, from a dystopian future that seems hundreds of years away. What if, however, these depictions were not far off of the realm of possibility within our lifetime? Could you imagine being hunted down by an incredibly well-proportioned individual that you couldn’t even tell was a killing machine? All jokes aside, this kind of robot that is indistinguishable from a human may not be as far-fetched as many currently think it is. Even those attributes of a humanoid robot which we may consider to be positive could in the long term damage us as a society in other ways we at this moment do not foresee. It is for this reason that we believe we must constantly be evaluating our choices in the development of such AI that resembles us, so that we may at least avoid the pitfalls of their creation, if not avoid their ultimate creation altogether.
One large concern that we have surrounding the development of human-like AI is the aspect of human interaction with the systems.
Another aspect of humanoid robots that we must consider is how the robots will be influenced by us as humans, and through this change then again affect our society. What we mean to say is that the biases t
Another aspect of humanoid robots that we must consider is how the robots will be influenced by us as humans, and through this change then again affect our society. What we mean to say is that the biases that humans have can even shape the way in which robots, these almighty, omniscient ‘beings’, view the world. Just as today the biases with which we operate are leaked into our machine learning models, these biases could be transferred to our humanoid compatriots. While, of course, in some ways this can be good, such as in terms of safety, being able to determine which animals to approach or which to stay away from, like a lion or a dog (while one may still be more dangerous than the other, however such is the nature of bias). Contrarily, the negative biases that humans have, which are much more visible in today’s media-based society, can also be transferred to these robots. In an experiment in ‘conversational understanding’, humans taught Microsoft’s AI chatbot to be racist and biggoted within a day. Could you imagine what the outcome would be of a humanlike AI that inherits the longstanding biases existent in our society for centuries? Perhaps this is a lesson that before we can endeavor in such feats we must fix our biases as humans first.
Humans Turning Into Cyborgs
While AI is becoming more and more human-like over time, humans are simultaneously becoming more equipped with AI as well. Some technological applications that have already been incorporated in humans comprise smart prosthetics that adapt to the wearing person’s missing limb, electronic skin that can register touch and can sense temperature differences or brain-computer interfaces that re-enable paralyzed patients regain control over their limbs. Implementations such as these have been invented to help people recover to the capabilities that they used to have or should have had to begin with. There’s no arguing the fact that a paralyzed or disabled person should deserve the chance to walk again. However nobody is stopping you from enhancing your own performance with AI.
Up to now, AI-enhanced humans do not outperform others for instance in sports or job activities. For the future however, this might not be the case. If we were to take the earlier mentioned brain-computer interface and further develop its abilities, it might not just move your limbs, it might also just increase your IQ by a significant amount. For instance by expanding your long-term memory or with pre-installed knowledge that is instantly obtained by its user. Anyone who would own such a device, would instantly be favored by corporations that seek out the most intelligent workers. When these sorts of applications would become commercially available, the end of the current state of equality would commence. AI-enhanced humans will dominate top positions and with their accumulated wealth, their descendants will be enriched with the same technology. This will create a certain brain-computer imperium where ‘cyborgs’ thrive and the simple unenhanced people will struggle to find a job that doesn’t solely consist of manual labor. This dystopian future is not at all hard to imagine and thereby stresses the urgency to prevent such outcomes before they’re a thing of the present.
The situation sketched above might sound rather alarming, however in a way this transition is already happening as we speak. With the incorporation of smartphones and other smart applications in our daily lives, we have already outsourced numerous responsibilities to our devices. For instance, just think of how many telephone numbers you have remembered at this moment and then how many of those you learned after your first smartphone. Chances are, this amount approaches zero. Another example is an Irish study where children with smartphones performed significantly worse on math and reading skills compared to their peers who did not own such devices. These examples might just be a direct result of us being able to consistently rely on applications that perform these tasks for us. If these possibilities are extended, this might come at the cost of our own capabilities, eventually dumbing down our population as a whole.
How Can We Stop This Process?
The technologies surrounding AI are moving towards the creation of humanoid robots incredibly quickly, so much so that at this point we would say it is inevitable that they will be among us soon. Taking this into account, it is up to us as a society to determine what their role will be in our lives, and how much impact they will have on us socially and personally. While this is something that people from all facets of our civilization will have a say in and be able to help determine, it is up to us as AI experts to facilitate the discussion concerning the possible issues. Our hope is that we will be able to spark a conversation that ignites our creative brains to think of ways in which we can responsibly introduce these machines into our society.
Regarding the enhancement of humans with AI applications, it is clear that with the development of these – as for now – medical applications, there is a whole area of yet unexplored opportunities of human enhancement that is on the brink of being discovered. If these appliances are employed to let disabled regain their physical capabilities, these advancements can only be applauded. However, with the imminent arrival of similar inventions for enhancing ‘standard’ human performance, we should enforce worldwide legislation that prevents technologically boosted humans from occupying all top positions.
It is obvious that it is no longer a question whether AI and humans are converging, but rather where it will stop. To quote economist Jim Womack:
“History has taught us not to bet against technology and technologists”Jim Womack, author of “The Machine that Changed the World”
AI will become more human-like and humans will be equipped with AI-enhancement. It is however up to us where we should draw the line.